Nanodots
The surface of a solid substance has different properties to the bulk solid which underlies it, sometimes significantly so. Atoms in the bulk are surrounded by other atoms from the substance itself; atoms at the surface are additionally exposed to the outside world, and may be able to react chemically with molecules in the surrounding gas phase. Common examples of these interfacial or chemisorption reactions include the rapid formation of a thin passivation layer of oxide on a freshly-cut sample of aluminium, or the hydroxy species derived from atmospheric water which are present on the surface of silicon dioxide.
Chemisorption involves some degree of charge transfer between the solid phase and the adsorbed species. As a result of this surface charge, if the solid is a semiconductor (or insulator) the energies of its conduction and valence bands are different in the region very close to the surface from their energies in the bulk. This phenomenon is termed band bending .
For example, tin dioxide is an n-type semiconductor. If oxygen is chemisorbed on its surface, it initially removes electrons from the conduction band, to form dioxygen anions at the surface, and a depletion layer in the tin dioxide just below the surface. If some of the surface oxygen reacts with gases such as hydrogen, this effect will be partially reversed.
If a junction is produced between two different semiconductors (for example, by diffusing two different impurities into a single silicon crystal), then it is well known that electrical conduction across the junction is non-ohmic due to the existence of a built-in potential. Similarly, when a sample of a single semiconductor has a morphology of distinct grains (for example, a compacted powder) the presence of surface species and the resulting band-bending will lead to a potential barrier between adjacent grains. Conductivity across these barriers is generally thermally activated, showing Arrhenius-type behaviour, and may dominate the overall measured conductivity of the sample.
Where thermal treatment has caused the grains to grow into each other to some extent (neck formation), different behaviour may be observed. As the grains are now interconnected, the existence of intergrain potential barriers cannot be assumed. However, chemisorption, and reactions of the chemisorbed species, can still produce or modify a depletion region close to the surface. If the 'necks' between grains are narrow, then the depletion layers can 'choke' the conductivity to varying extents.
We may now be able to see how a granular sample of a semiconductor can act as a gas sensor:
(i) the surface of the grains comes to equilibrium with the gas environment, with gas molecules physisorbed, chemisorbed, or both
(ii) charge transfer involved in chemisorption leads to band bending and modification of depletion layer in each semiconductor particle
(iii) the electronic effects in (ii) produce either a potential barrier between the particles; or a variation in the depth of a depletion layer within intergrain necks
(iv) thus overall conductivity of the sample depends on the nature and extent of its interactions with its gaseous environment.
(Note that some semiconductor gas sensors actually involve interactions between gas molecules and the bulk semiconductor, but these will not be considered here.)
As semiconductor gas sensors of the type described here work through gas-surface interactions, it is likely that the smaller the semiconductor particle, the more efficient the sensor. For a particle of radius r , the surface:volume ratio is proportional to 1/r. In other words, a small particle is "all surface". This is borne out by experiment. For example, Ansari and co-workers made hydrogen sensors from films of tin dioxide particles of 20 to 50 nm diameter. They found that the sensitivity increased rapidly when diameter was less than 25 nm. (See: R. C. Aiyer, Z. A. Ansari and R. N. Karekar, Thin Solid Films, 295 271 (1997).)
As it is possible to produce nanoparticles with the same dimensions as typical depletion layer depths, it may be possible to deplete a particle entirely as a result of interactions with its gaseous environment (or restore all charge carriers by interaction with a different test gas). Such a loss / restoration of all mobile charge carriers should have dramatic effects on the overall conductivity of a sample. Also, as we intend to produce electrodes of similar dimensions, we hope to be able to probe the conductivity of single nanodots.
Typical synthetic routes to compound semiconductor nanodots involve thermolysis of molecular precursors in organic solvents such as tri-n -octylphosphine (TOPO) or 4-ethylpyridine (the former can also act as a source of phosphorus in the synthesis of InP, etc). For example:
The 'dots' so produced are usually coated with a layer of the solvent or surfactant used. We aim to change the surface derivatisation of the dots so that any surface coating is easily removed by low temperature processing. A further aim is to keep reaction temperatures as low as possible, to avoid aggregation of the dots.